Mechanistic approach exploring combination agents to potentiate anti-leukemia activity of Hu8F4, a clinical stage T cell receptor mimic antibody Free

Hu8F4 is a T cell receptor (TCR)-mimic antibody that binds the leukemia-associated PR1/HLA-A2 complex overexpressed on HLA-A2+ leukemia cells. In a phase I dose escalation trial of Hu8F4 in patients with relapsed/refractory AML, MDS, and CMML, peripheral blast reduction following Hu8F4 infusion suggested biological activity and therapeutic potential. The initial clinical data and tolerability of Hu8F4 motivated us to explore approaches to improve the efficacy of Hu8F4 by testing it in combination with agents involved in immune checkpoint and/or phagocytosis.

We previously identified antibody-dependent cellular phagocytosis (ADCP) as a principal mechanism of Hu8F4 anti-AML activity in vivo. We showed that impeding “don't eat me” signaling of CD47 via blocking with anti-CD47 antibody or knocking out CD47 via CRISPR/Cas9 significantly improved activity of Hu8F4. Because of the challenges in developing anti-CD47 antibodies in clinic, we chose to assess potential synergy of other phagocytosis regulators with Hu8F4.

We first tested whether blocking SIRPα, a CD47 ligand on macrophages (MΦ), led to an increase in phagocytosis, as we observed with anti-CD47 blockade. To assess this, we used THP1 (endogenously HLA-A2+) and U937-A2 (HLA-A2-transfected) leukemia cell lines as targets and NSG mouse bone marrow derived macrophages (BMDM) as effectors at an initial E:T=1 in ADCP assay. While Hu8F4 alone was insufficient to prevent growth of target cells, adding anti-mouse SIRPα (blocking clone P84) slowed by 92% the growth of THP1 cells at a concentration 10 μg/ml, but did not affect growth of the more aggressive U937-A2 cells even at high concentrations (100 μg/ml) of P84. Given that anti-CD47 F(ab)'₂ combined with Hu8F4 slowed growth and eliminated both target cell lines by day 5, we concluded that blocking SIRPαwas less effective than blocking CD47. To test if SIRPα blockade could improve the efficacy in Hu8F4 in vivo, we injected NGS mice with U937-A2 (5x10^3). Mice were then treated with Hu8F4 +/- anti-SIRPαantibody (1mpK three times per week for a total of 10 doses starting on day 5; n=5 per group). Control mice and mice treated with anti-SIRPαantibody alone both succumbed to leukemia by day 30. Single-agent Hu8F4 treatment slowed leukemia growth and improved survival as we have shown previously, but adding anti-SIRPα treatment did not improve survival, consistent with the in vitro result. It's possible that saturation of SIRPα was not achieved in the mouse models since it is highly expressed on mouse MΦ. Higher amounts of antibody may adequately test the full effects of SIRPα blockade. We also tested recombinant SIRPα-Fc (TTI-621), another agent in clinical development designed to block CD47. However, neither TTI-621 alone, nor in combination with Hu8F4, increased ADCP.

Macrophages, key actors in ADCP, play dual roles in the tumor microenvironment: MΦ can remove tumor cells whereas MΦ that express TREM2 (Triggering Receptor Expressed on Myeloid cells-2) may be immunosuppressive, thus supporting tumor growth. TREM2 deficiency, or blocking with anti-TREM2 antibody (Clone178), has been shown to remodel MΦ in the tumor microenvironment to enhance anti-PD1 immunotherapy (Molgora, Cell, 2020). Thus, we differentiated NSG bone marrow cells into MΦ in presence of anti-TREM2 antibody (178), to stimulate antitumor activity. However, we saw no improvement in ADCP activity of Hu8F4 in our in vitro system when anti-TREM2 was used. This luck of efficacy is likely due to lower expression of TREM2 on NSG-BMDM, compared with BMDM derived from C57BL/6J mice, used in the study by Molgora et al.

Low-density lipoprotein receptor-related protein (LRP1) is highly expressed on MΦ. It has been implicated in MΦ M2 polarization and has been linked to PD-1/PDL-1 therapy resistance in patients with solid tumors. To test if LRP1 signaling plays a role in Hu8F4-induced ADCP we used antibodies to LRP1 developed in our laboratory. In pilot experiments anti-LRP1 antibodies improved ADCP effect of Hu8F4 by up to 80% (p<0.0001) in the presence of NSG BMDM, however the mechanism and in vivo activity of a-LRP1 antibodies are under investigation.

In summary, we tested both well-known and novel regulators of ADCP, such as SIRPα, TREM2 and LRP1 in combination with Hu8F4. Understanding the mechanisms and potential contribution of these agents has implications not only for Hu8F4, but for other TCR-mimics and other antibodies that work through ADCP as well.